| title | RoAR-VAE: Improving Latent Diffusability through Robust Prediction |
|---|---|
| authors | Yang Yue, Fangyun Wei, Yitong Wang, Jinjing Zhao, Henry Zheng, Dong Chen, Gao Huang |
| date | 2026-10-19 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2610.26216 |
| affiliation | Tsinghua-LeapLab |
| upvotes | 40 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf code |
| abstract | An effective visual tokenizer must preserve image fidelity while producing latents that downstream diffusion models can learn efficiently. Prior work improves latent diffusability through semantic alignment or joint tokenizer--diffusion training. Revisiting REPA-E, a representative approach, we find that removing its diffusion objective and dedicated generative blocks while retaining semantic prediction from corrupted latents yields comparable downstream generation performance under a matched setting. This observation motivates a general principle of robust prediction: generation-relevant information should remain recoverable from corrupted latents. We instantiate this principle with two complementary objectives. RobustAlign predicts pretrained semantic features from corrupted latents to preserve high-level structure, while RobustRecon reconstructs the clean image to retain fine-grained spatial and appearance information. Together, they form Robust Alignment and Reconstruction VAE (RoAR-VAE). Our analysis suggests that robust prediction improves diffusability by making useful information more accessible in high-variance latent directions, which diffusion models tend to learn earlier. Empirically, RoAR-VAE improves average generation benchmark performance by 2.19 points over the strongest baseline tokenizer under the same training budget, while improving small-text reconstruction accuracy by 16.93 percentage points over FLUX.1-VAE. These results demonstrate that RoAR-VAE achieves a favorable balance between reconstruction fidelity and downstream generative learnability. |
cite
@misc{yue2026roar,
title={RoAR-VAE: Improving Latent Diffusability through Robust Prediction},
author={Yang Yue and Fangyun Wei and Yitong Wang and Jinjing Zhao and Henry Zheng and Dong Chen and Gao Huang},
year={2026},
eprint={2610.26216},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2610.26216}
}